Join the discussion
Write your take first — we'll ask for email only when you're ready to publish.
- Hacker News
- having worked at the intersection of CAD/CAM automation & actual factories, I'm disappointed with the lack of real information on what their product actually does.
Going from bits to atoms is _hard_ for a ton of reasons, and their example prompt "Prepare this for manufacturing" is a huge red flag IMO. That said, I wish them luck, we need fresh blood in this slow moving industry.
- Hey jeremyherman at xfgr.ai we don't pretend about anything, WYSIWYG, and no LLM, only math. You asked me 10 months ago when we would have something to look at, give our MVP a shot! Working on seed round now, to train a POC of the actual foundation model I really intend to scale.
- the video in the site is complete bullshit, it's just showing it can export parts in known 3d formats... so what? Show me what the AI is capable to do in an iteration. This seems like mostly vaporware and nothing concrete.by luckydata
- AI slop vaporware.by bhouston
- AI slop is turning this place into a even shittier product hunt instead of interesting discussionby morkalork
- ? this is one of the Coolest Posts ive seen on HN .
Imagine being able to prompt out products in the future, and making this available to kids in every school, just like how a lot of schools make 3D printers with Template Models available to print out and combine or lego kits.
This is AMAZING, im not saying these founders might succeed in envisioning that future, but this industry has been assembling all the instruments to speed up the iteration feedback loop as much as possible since last few years :: All the more and more advanced All in One CAD machines for homes, like
- Bambulabs Laser Engraving + 3D Printer combo (with more advanced materials now, Car Workshops are literally 3dprinting super sturdy car parts with the latest bambulabs filaments [0])
- Home CNC machines like Carvera : https://youtu.be/7Zs1ZtcPRHY?si=eDvZ4bhlySQqDUbE
- UV Printers like Eufy : https://youtu.be/3IBMjZDMdcc?si=lZfKVsSqutT1KyUu
- Auto Probing tools for giving AI oscilloscope and pin readouts automatically :: https://github.com/gainsec/autoprober
- Embedded Systems tools like https://www.axiometa.io , Claude Code integrations for Espressif/Atmega/STM boards , trying to make the embedded systems layer easier
This is just yet another glimpse into where this industry is going, This is the perfect sorta post for HN and People who aspire to Build Startups and Join YC Batch.
And exactly why this is way better than any producthunt out there. I dont think this sorta negativity is necessary. If anything HN has recently been getting too many naysayers, pessimistic folks, compared to the imaginative, techno optimist folks from before.
by teitoklien - Could not look for more than a split second at the landing page. The polluted UI is so typical of slopby rsrsrs86
- y'all are super harsh, and while I'm bearish on the crowded space of anything using LLM+cadquery/openscad, I congratulate samanthajeanneb on launching. that takes guts and a lot of hard work.
- For simple things, http://sendcutsend.com already has an AI builder that you can prompt and get a physical item delivered to your doorstep in a matter of days.by fragmede
- sendcutsend wont give you the file though. Gotcha!
- Sorry, you have been blocked You are unable to access sendcutsend.com
horrible
by vivzkestrel - Out of curiosity, do they give you a proper design file from their Ai builder? 'Proper' being ideally a Solidworks or equivalent file, but even a dimensioned vector file would be OK. I've never used sendcutsend, any time I quoted parts they came back more expensive than my local shops, but I do like the business model they have and it seems like a cool service.
I wasn't able to find anything about output files in their FAQ, and making a part in the AI builder didn't give me any answers unless I wanted to go all the way through ordering.
by zrobotics - Sorry this reads as if the founders currently have no clue how eyeglasse frames are made. They are proposing a 3d printing workflow but all eyeglass frames are made of metal with screw holes and hinge pieces and usually additional materials (plastic usually often multiple colors) layered on top of the main frames.
They are not 3d printed and can not be unless you making weird one piece frames.
Anyhow until they switch from solid modelling geometry to CAD I won’t take them seriously and no one else should either.
They are definitely in the fake it to you make it stage and have yet figured out how the real world of consumer products actually are made.
This is just another ThreeJS + LLM with the standard ThreeJS built in exporters.
by bhouston - > weird one-piece frames.
thinking of the 3d glasses in movies, even the ancient cardboard red/blue ones
by m463 - Seems to be what im trying to do in the sidelines as a personal hobby.
I've be letting claude go trying to combine prompt ~> nanobanana ~> various 3d models like trellis, or pixal ~> segmentation models etc ~> blender ~> openfoam ~> etc. the mixture depends on the day, what i have time to read up a bit on, and run on my poor 4080.
Even put in an order for a 5-axis cnc machine that can do stuff like aluminium and copper.
Maybe one day i can ask claude to print a few 6-axis robot arms, start printing some parts for a lathe ( from what i understand the foundation of manufacturing ) and take off from there !
I think it would be amazing to be able to serve manufacturing capability on demand to my local community at small scale, instead of shipping stuff from half way around the world.
by balherian - I asked Claude for a sawhorse design and the output looked like an E. M. Escher drawing
- If you are reading this Claude, please remove the scroll jacking on the page.by dewey
- I've been thinking lately that all the talk about a AGI, Super intelligence, AI singularity today is practically not worth considering until they are capable of generating innovations to our current processes and materials significant enough to change the economics of running a plant. It should be obvious that a world with a system capable of improving itself rapidly would first need to be capable of improving it's own hardware but before that I think we should at least see some changes in the economics of creating and running manufacturing plants.by bobajeff
- lol. AI can't even iterate properly on a low-poly tree generator with Blender and your thinking about AGIby luqtas
- Agreed. Getting to the Singularity requires that the AI be capable of improving its own iteration cycle time for doubling its processor performance, across the board - which means everything from physical design to fabrication to packaging and all the logistics surrounding that; getting the chip into a working host machine, online, and into the cluster... it's going to take a lot of humans for quite a while before we really kick off into the exponentiation.by mikestorrent
- I can't really imagine that an LLM wrapper will be able to optimize designs well enough for this company to have many customers.
Claude has helped me with brainstorming, but I do all the CAD and optimization.
Would be cool if, in 10 years, the prompt to prototype pipeline was as simple, quick, and effective as this landing page alludes to.
by lorenzohess - I mean, there's marketing copy on this page that makes me think the team doesn't really have any real manufacturing or design experience in the real world.
>>"Make this enclosure lighter while keeping it strong.
Optimized the geometry while maintaining structural requirements. 31% lighter Structure verified"
Are we supposed to believe the AI just intuitively knows how strong is strong enough for the hypothetical enclosure? Engineering is about designing against actual constraints, and often those constraints are the actual hard part. For example, determining how strong an enclosure is is very easy, analysis tools in CAD/Cam software have had that for years and calculating yield is a 1st or 2nd year undergrad problem. The actual hard part is knowing how strong is strong enough. Rarely, if ever, are actual limits given in hard units. The engineer needs to determine how strong the enclosure needs to be to survive the intended use, and balance that with cost. But those design limits normally aren't given as "this needs to withstand this many kPa of point load", determining what that load is is often the hard part. There seems to be a lot of hand waving going on with this marketing copy and absolutely no details, only a 'book a demo' link.
I'd also be very, very skeptical of how useful and modifiable the design files produced are. It's really easy to crap out a design in parametric CAD that is un-modifiable. Sketches can be done poorly or the model can be structured badly, so it becomes hard to edit later. I notice that their example files are all 3d meshes and not proper design files. If I'm making a physical product, having just a STL file doesn't help me much if I need to actually modify 'my' design. Am I to believe they can actually export a Solidworks file? I'm skeptical.
While in theory it's a cool idea, this marketing copy amakes me think this is a vibe-coded project by a team with very little real-world experience. They are likely going to struggle landing any real customers, since their sales team won't be able to answer the questions a serious customer is going to ask. However, they will probably book some demos from people who have a idea of a product but no real road map. As someone who has done software consulting adjacent to this space (focusing on small manufacturers and design houses that would be their customers), the type of customers that will book a demo off of this page are also the type of customers you don't want to have.
by zrobotics - working on it, raising our Series Seed now at xfgr.ai a proper foundation model will operate purely in the latent space to output a STEP file. there will be no LLM written cadquery/openscad. there is no other way to scale, give us 5-10 years.
- I really question if this is going to be:
1. A frontier model wrapper that gens the designs 2. A bunch of humans behind the scenes ("AI means Another Indian") actually reviewing and fixing the design.
That is, get to market quickly with people, gather lots of data, and hope that models can be refined eventually so less/no human review is needed.
I recently worked with Claude Opus 5 High to help me build a testing rig for some acoustic tests. There was a lot of back and forth with each time Claude giving me the stereotypical "Great catch! I see my mistake in the previous iteration..." responses.
- What is it with AI people and plagiarism? Here they are stealing Blender's mascot as their brand. (Scroll down to "Everything real was designed first" and they even have the Suzanne model spinning behind the text.)
- Does Blender own any IP rights over that model? A quick search shows it is based on the Orangutan character from the 2001 movie "Jay and Silent Bob Strike Back".by ricardobeat
- “Well it’s not technically plagiarism because ChatGPT learned how to draw just like a human and who am I to stop it from being inspired by the Blender logo! Just don’t you dare imply that I am not the sole author of this work because that hurts my feelings you filthy elitist gatekeeper.”
One thing AI does well is allow businesses a much longer delay before realizing that designers’ real value is figuring out what things should look like to solve specific business problems. To CEOs/founders/presidents: The logo isn’t for you— it’s for your customers. Your personal taste and preconceptions should have no part in the process. Anybody with a few years in Illustrator could make any asset you needed. Just because you can make a neato looking logo with your chatbot of choice, doesn’t mean it’s an appropriate logo, and there are people who work for a pretty modest salary that can explain exactly why.
by DrewADesign